KAPILASHRAMI, E., SAHAJWALLA, V., and SEETHARAMAN, S.

Slag formation during high-temperature interactions between SiO2-containing

refractories and iron melts with oxygen. VII International Conference on Molten Slags Fluxes and Salts, The South African Institute of Mining and
Metallurgy, 2004.

Slag formation during high-temperature interactions between

SiO2-containing refractories and iron melts with oxygen
E. KAPILASHRAMI*, V. SAHAJWALLA, and S. SEETHARAMAN*
*Royal Institute of Technology, Stockholm, Sweden
School of Materials Science and Engineering, The University of New South Wales, Australia

Refractory/metal interactions and associated reactions are significant during casting as they can
lead to inclusion formation. High-temperature interactions between aluminosilicate refractories
and molten steel were studied in the present investigation using the sessile drop method. The
changes in the contact angles between molten iron with well-defined oxygen levels and substrates
of silica, mullite (3Al2O32SiO2) as well as commercial refractories were measured in the dynamic
mode. In all cases, the reaction led to slag formation.
The experiments were performed under isothermal conditions at 1823 and 1873 K. A known
and constant oxygen partial pressure was imposed above the iron drop, through appropriate CO-
CO2-Ar mixtures. For all substrates, the contact angles started decreasing due to the lowering of
the surface tension of iron, as oxygen was imposed on to the system. When a critical level of
oxygen was reached, slag started forming at the drop/substrate interface and at this stage, the
contact angle dropped suddenly. Subsequent behaviour of silica and mullite was observed to be
different, with SiO2 showing a non-wetting behaviour. In the case of SiO2 substrate, there were
also deep erosions along the periphery of the drops and there was slag accumulation at the sides of
the metal drop, probably due to Marangoni flow and/or SiO(g) formation. SEM and EDS analysis
of the substrates revealed that the formed slag was fayalite.
The thermodynamic criteria for slag formation and possible mechanisms of the reaction are
discussed in the light of the experiments.
Key words: silica, mullite, fayalite, corundum, slag formation, sessile drop.

Introduction it is crucial to have an understanding of the

The present work was initiated aiming at an understanding
of the reactions between silica containing refractories and
liquid iron containing well-defined amounts of oxygen. The
work is of relevance to clean steel production, as the
refractory erosion products can be sources of non-metallic
inclusions in the steel. Erosion of refractories by steel with
slightly higher amounts of oxygen, during the tapping of
LD-furnace and erosion of the refractory runners during
ingot casting, are the process steps where the present study
is of relevance. As the refractories commonly in use are
aluminosilicates, the present studies of the reactions
between oxygen-containing molten iron and SiO2 as well as
mullite were initiated.
The reactions between the above-mentioned refractories
and oxygen-containing iron are characterized by the
formation of molten slag at the interface. The slag
formation seriously affects the mechanisms of the
refractory reactions and thereby has serious implications on
the performance and the life of the refractories. Thus, the
present paper aims at an understanding of the implications
of slag formation for refractory life from a fundamental
view point. The studies were carried out in micro scale by
sessile drop technique, whereby, the reaction was followed
by visual observation of the drop images dynamically.
thermodynamics of the reactions occurring between the
refractory oxides and iron containing oxygen as well as
with pure iron. As part of the present work, a
thermodynamic study was carried out where the stabilities
of the various phases in the system Fe-O-Si were calculated
using thermodynamic data in literature. From this input
data, a phase stability diagram, presented in Figure 1, was
constructed. The calculations showed that the two silicates
likely to be formed in this system are orthosilicate and
metasilicate. However, at steelmaking temperatures, only
the orthosilicate or fayalite in liquid form will be the stable
phase. On the other hand, as the system is cooled down, the
metasilicate can precipitate. The constructed phase stability
diagram shows that Fe2SiO4 would be formed if the oxygen
content in the molten iron exceeded 0.1 wt%. This is likely
to happen during certain steelmaking steps where the
oxygen potential in the slag is high.
In the case of aluminosilicates, the reactions are a bit
more complex. An examination of the phase diagram1 of
the system Al 2O 3-SiO 2 reveals the formation of a very
stable compound, mullite, with the formula 3Al2O32SiO2 as
well as 2Al2O3SiO2. The reaction and its Gibbs energy of
formation2 for this compound are as follows:

Thermodynamic background 3 Al2O3 + 2 SiO2 = 3 Al2O3 2 SiO2 [1]

To understand the nature of the reactions of slag formation, G1o = 18.779 26.953 T J mole 1

SLAG FORMATION DURING HIGH-TEMPERATURE INTERACTIONS 417


Figure 1. The phase stability diagram for the system Fe-O-si
As seen, the Gibbs energy of formation is very low and it
is therefore reasonable to assume that this could be a
predominant phase in the runner refractory. The reaction
between molten iron containing oxygen and mullite could
result in the formation of fayalite slag and hercynite
(FeAl2O4) particles. The reaction and the Gibbs energy of
formation2 for fayalite are:
2 Fe + O2 + SiO2 = Fe2 SiO4
G2o = 518.959 + 116.82 T J mole 1 (T range 1000 1400 K )
Figure 2. Phase diagram of the system FeO-Al2O3-SiO2
418
At temperatures above the melting point of iron, the silica
part of mullite will react with the metal and form fayalite.
As mentioned previously, the ironmetasilicate being stable
at lower temperatures would not be present in the system.
The Gibbs energy of formation of solid hercynite, FeAl2O4,
is given by the equation2:
1 [3]
Fe + Al2O3 + 2 O2 = FeAl2O4
G3o = 328.349 + 82.044 T J mole 1
The phase stability diagram of the Fe-Al-O system at the
steelmaking temperature has been presented earlier by the
authors. 3. When solid hercynite and liquid fayelite are
formed, there is likely to be a reaction between these two
and they will form a ternary slag that will have a
composition that changes with temperature, according to
the phase diagram FeO-Al2O3-SiO24 presented in Figure 2.
The formation and composition of this slag will not just be
dependent on the thermodynamics of the reactions but also
on the kinetics of the reaction.
Dynamic investigationexperimental work
In the present work, the reaction between silica containing
substrate and iron drop containing oxygen were
investigated on a micro level by means of the Sessile drop
technique. The contact angle between the substrates and
iron containing varying amounts of oxygen were
[2]
investigated in static as well as in dynamic mode. The
MOLTEN SLAGS FLUXES AND SALTS
experiments were carried out in purified argon as well as complemented by the image analysis part developed at the
with varying oxygen partial pressures imposed on the Division of Metallurgy, Department of Material Science
system by means of suitable mixtures of CO, CO2 and Ar and Engineering, Royal Institute of Technology, Sweden.
into the system. Details regarding the furnace, the gas
cleaning system as well as the image recording device have
been presented elsewhere56. The change in contact angle
Results and discussions
was followed as a function of time at different The experiments were carried out in purified argon gas as
temperatures. well as with different partial pressures of oxygen, the
The samples were heated up in purified argon gas and the highest partial pressure corresponding to 0.19 wt% O at
appropriate Ar-CO-CO2 gas mixture was then introduced 1823 K and 0.29 wt% O as oxygen dissolved in molten iron
into the system after the attainment of the experimental at 1833 K, i.e. more than required to form fayelite slag
temperature. The X-ray images of the drop along with the according to the thermodynamic analysis.
substrate were taken in the video mode, following the
changes in contact angle with time. The experimental Results with silica substrate
details, the temperatures, gas ratios and their corresponding Images of the drops from the experimental series with the
values of dissolved oxygen in molten iron, are presented in highest oxygen partial pressure and at 1833 K are presented
Table I. At the end of the measurements, the Ar-CO-CO2 in Figure 3. It is seen that the drop flattens as the
gas mixture was replaced by purified argon gas and the atmosphere is changed to the gas mixture with higher
sample was then quenched. The substrate was taken out and oxygen partial pressure confirming, thereby, that the
the cross section was examined by Scanning Electron contact angle decreases due to the introduction of oxygen,
Microscope (SEM) and also through Energy Dispersive which is surface active. An interesting observation in the
Spectroscopy (EDS). Some selected experiments were then present series is that the contact angle exhibits a tendency
repeated in order to confirm the reproducibility of the to increase after reaching a minimum point.
results. A reasonable explanation for the initial decrease in the
The contact angle between the substrates and iron drop contact angle would be due to the increase in the surface
was evaluated by the video film frames obtained using a oxygen concentration by oxygen pick up from the gas
software, developed at the Department of Material Science phase. The oxygen from the surface has to diffuse into the
and Engineering, Carnegie Mellon University, USA, bulk which, perhaps, would account for the plateau region.
Further decrease in the contact angle is indicative of the
attainment of the equilibrium oxygen concentration at the
Table I
metal-substrate interface and the progress of the reaction.
Experimental sequences and their conditions At this point, the drop would also have reached the
equilibrium oxygen content with respect to the gas phase
Experimental CO/CO2 Targeted Corresponding Temperature
and enough oxygen will be present in the system for
series ratio PO2 Pa ppm O (K) fayalite slag to form. There is a likelihood of a thin fayalite
layer formation between the substrate and the drop, which
A - < 10-14 - 1823.1873 would arrest the reaction. At this stage, the contact angle
B 0.28/0.42 1.510-2 2900 1873 measured would, most probably, be between the iron drop
B 0.28/0.42 1.510-2 1960 1823
with oxygen potential equal to that of the gas phase and the
C 0.42/0.28 3.010-3 871 1823
D 0.56/0.14 9.910-4 327 1823
thin slag film on the silica substrate. This, evidently, is the
fayalite slag formed by the following reaction:

1 second 10 seconds 20 seconds

36 seconds 38 seconds 52 seconds

Figure 3. X-ray images of the sessile-drop at 1833 K at different time interval as an oxygen partial pressure of
1.510-2 Pa was imposed at zero time

SLAG FORMATION DURING HIGH-TEMPERATURE INTERACTIONS 419

 2 Fe( l ) + 2 O( Fe, l ) + SiO2 ( s ) = Fe2 SiO4 ( l ) [4]
Beyond this point, the iron drop showed a tendency to
move from its original position. This behaviour is likely to
be due to the imbalance in the forces acting on the drop. A
similar phenomenon was observed by the present authors in
the case of the dynamic contact angle measurements
between liquid iron and alumina substrate3.
The SEM micrograph of the reacted drop with the
substrate after the experiment at 1833 K is presented in
Figure 4. The insert in the Figure shows the location of the
field of vision vis--vis the iron drop. The low contact angle
between the drop and the substrate is quite evident in this
Figure. It is also seen that a thick product layer is formed by
the side of the drop as mentioned earlier. The atomic ratio
of iron to silicon in this layer, obtained by Electron
Dispersion Spectroscopy (EDS) was close to 2, confirming
that this phase was indeed the fayalite slag. The figure, on
higher magnification, showed clearly the formation of
dendrites in the slag. These were found to be pure SiO2 by
EDS analysis. The SEM micrographs of the drops at other
oxygen partial pressures and temperatures were found to be
quite similar, the amount of slag decreasing with decreasing
oxygen partial pressure.
In some of the experiments, there was a preferential
corrosion observed along the periphery of the drop. This
may be caused by the formation and evaporation of SiO,
either due to direct decomposition of SiO2 substrate at the
low oxygen potentials prevailing in the gas phase or the
reaction between silicon dissolved in the liquid metal and
SiO2 substrate, both the reactions leading to the formation
Iron drop
Figure 4. Photomicrograph of a section of the reacted drop and
the silica substrate after the sessile-drop experiments at 1833 and
oxygen partial pressure, PO2 = 1.510-2 Pa
420
of SiO gas. These reactions would be dampened as the slag
layer was formed around the bottom of the metal drop.
Another plausible explanation is that the erosion is caused
by Marangoni flow in the melt induced by differential
surface tension due to concentration gradients.
Results with Mullite Substrate
The results observed in the case of mullite, viz. the
immediate decrease in contact angle with the introduction
of the gas mixture with high oxygen potentials, the
somewhat steady state when oxygen diffuses from the
surface of the drop to the bulk and then the final slow drop
in the contact angle were similar to those of silica substrate.
However, the slag formed during the reaction between
mullite and molten iron would be different from the silica
system. The thermodynamic analysis shows that when
mullite and iron containing some amounts of oxygen starts
reacting, fayalite and hercynite will be formed. According
to the ternary phase diagram presented in Figure 2, the
mullite, fayalite and hercynite will start to dissolve into
each other and form a slag. The composition path of the
slag will go towards mullite and an intermediate phase, iron
cordierite could be formed. At the experimental
temperature, the ternary slag would consist of FeO-Al2O3-
SiO2, the final composition being marked as x in Figure 2.
If the reactions are allowed to continue, this slag will be in
equilibrium with mullite and corundum. This is in
agreement with the results of the SEM-EDS analyses.
However, due to the kinetics of the reactions, there could
still be particles of iron cordierite and hercynite as the
reactions may not reach equilibrium. The composition of
the slag would also change during cooling and low
temperature phases could precipitate and this makes the
EDS analysis of the slag very complicated and difficult.
The SEM micrograph of the reacted drop with the
substrate after the experiment at 1823 K and the highest
partial pressure of oxygen is presented in Figure 5. Once
again, the insert in the figure shows the location of the field
of vision vis--vis the iron drop. It is seen that the drop has
reacted with the substrate and has also sunk into it. It is also
seen that a molten phase has been formed and that there are
free particles in this phase. The slag phase, in the figure
entrapped by the molten iron, was magnified and examined
Figure 5. Photomicrograph of a section of the reacted drop and
the mullite substrata after the sessile-drop experiments at 1823
and oxygen partial pressure, PO2 = 1.510-2 Pa
MOLTEN SLAGS FLUXES AND SALTS
 SE, 255 Tetra, 255
AlKa, 18 SiKa, 6
Figure 6. EDS mapping of the slag formed during reaction between mullite substrate and molten iron containing oxygen
through EDS mapping and the result is presented in
Figure 6. It is seen that the slag mainly consist of alumina,
silica and iron oxide. The particles are found to be alumina
particles and at the periphery of these particles there is a
layer containing aluminium, oxygen and iron. This layer
was found to be hercynite. The EDS mapping also shows
that the composition of the slag varies.
The present experimental work also showed that other
factors, besides the composition of the substrate, will have
an impact on the reaction mechanism. Factors like
inhomogeneity, surface roughness and amounts of pores
will also be of importance. The analysis of the porous
mullite substrate, revealed that, along with the surface
reaction with the substrate, reactions were taking place
downwards into the substrate, which could escalate the slag
formation and corrosion of refractories.
The present work is of great importance for the steel
industry, especially for the steel plants targeting the
production of clean steel. The results show that the
composition of the refractory plays a great role in corrosion
of the furnace linings resulting in slag formation which, in
turn, can result in the formation of non-metallic inclusion in
the steel matrix. The present work shows that silica, even
though chemically bonded strongly in mullite, reacts with
molten iron that contains some amounts of oxygen and
form a slag. The formation of a thin, dense, non-wetting
hercynite layer between molten iron containing oxygen and
alumina substrates shows a protective influence of the
presence of higher alumina contents in the lining material,
leading to less reaction with steel and thereby resulting in
fewer non-metallic inclusions. A distinct possibility to
increase the life of the furnace linings is to coat them with a
non-wetting layer, for example, hercynite. This layer would
act as a barrier between the refractory and the metal and it
would prevent further reactions with the refractory. Of
course, the economic gain in clean steel with these changes
should be compared to the higher cost of refractory
production.
SLAG FORMATION DURING HIGH-TEMPERATURE INTERACTIONS
Oka, 255
FeKa, 19
Summary
The present work was focused on studying the slag
formation phenomenon during the reactions between
oxygen-containing molten iron and aluminosilicate lining
materials by means of X-ray sessile drop method. In the
case of silica or mullite refractories, the reactions resulted
in the formation of a slag. With silica substrate, the slag
was found to be fayalite (2FeOSiO2). In the case of mullite,
the slag phase formed contained alumina. The slag
formation led to non-wetting conditions and the movement
of the iron drop. Preferential corrosion of the silica
refractory due to possible SiO evaporation or Marangoni
convection was also observed in the present study. The
factors that influence the corrosion of lining materials, apart
from the chemical reactions, were oxygen content in the
metal, porosity and surface roughness of the refractories.
The present studies are expected to lead to the development
of new generation refractories, which are resistant to
chemical attack by molten steel, thereby leading to lesser
contamination of the steel produced by non-metallic
inclusions.
Acknowledgements
Part of the work was carried out at in the University of New
South Wales, Australia. Financial supports for travel to
Australia from Brinell Centre, KTH and Axel Hultgrens
Trust are gratefully acknowledged.
References
1. BOWEN, N.L. and GREIG, J.W. J. Am. Soc., 1924, 7,
p. 242.
2. BARIN, I. Thermodynamical Data of Pure
Substances, Part I and II, VCH, 1993.
3. KAPILASHRAMI, E., JAKOBSON, A., LAHIRI,
421
 A.K., and SEETHARAMAN S. Studies of the
Wetting Characteristics of Liquid Iron on Dense
Alumina by X-Ray Sessile Drop Technique. Metall.
Mater. Trans. B, vol. 34B, 2003. pp. 193199.
4. LEVIN, E.M., ROBBINS, C.R., and MCMURDIE,
H.F. Phase Diagrams for Ceramists, Reser, M.K (ed.).
Ohio, The American Ceramic Society, 1964, p. 696.
5. KAPILASHRAMI, E., LAHIRI, A.K., CRAMB,
422
A.W., and SEETHARAMAN, S. Investigation of the
reactions between oxygen-containing iron and SiO2
substrate by X-ray sessile drop technique. Accepted
for publication in Metall. Mater. Trans. B 2003
6. KAPILASHRAMI, E., SAHAJWALLA, V., and
SEETHARAMAN, S. Investigation of the wetting
characteristics of liquid iron on mullite by sessile drop
technique, to be published.
MOLTEN SLAGS FLUXES AND SALTS